"Michael S. Tsirkin" <mst@xxxxxxxxxx> writes:
> On Thu, Jun 20, 2013 at 02:41:37PM +0200, Thomas Rast wrote:
>> "Michael S. Tsirkin" <mst@xxxxxxxxxx> writes:
>>
>> > Verify that author name is not duplicated if it matches
>> > sender, even if it is in utf8.
>>
>> Small nit: if you make two patches out of it, add the tests first with
>> test_expect_failure. Then flip it to test_expect_success in the actual
>> code change. That makes it easy to verify that the test actually checks
>> the right thing, and that it was your code change that fixed it.
>
> I did this by reverting 1/2 and rerunning.
>
> But applying in reverse order means bisect can give us
> a setup where some tests fail, I thought it's a
> good idea to avoid that.
That's why you need to test_expect_*failure* in the commit that adds the
tests -- essentially saying "I know this is broken!".
Yes, it's a roundabout way. But splitting code and tests in the way you
just posted is equally roundabout, while not having the benefit that one
can check out the commit at patch 1 and verify that it is indeed broken
(showing up as "still have known breakage").
--
Thomas Rast
trast@{inf,student}.ethz.ch
--
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html